Expansion revenue should be the most predictable revenue in your business. Existing customers. Proven delivery. Demonstrated willingness to pay. You have their history, you know what they bought, and you know what happened after they bought it.
Proven delivery is a condition in that list, not a description. Delivery-failure churn precludes expansion. If you aren't delivering what you already sold, that's the work, and nothing below applies until it's done.
Instead it feels random and magical. When it happens.
Meanwhile the revenue that comes from people who have never bought anything from you, whose budget you're guessing at and whose timing you can't see, is the one everybody calls predictable. It gets a spreadsheet, a weekly review, and a number on the wall.
Something is backwards here. Let's be precise about what.
What "Predictable" Means When a Sales Team Says It
Watch how the number actually gets made.
Someone decides the company needs $8 million in new ARR. That decision comes first, and it comes out of a board conversation or a raise or a model that needed to end in a particular place. Then the funnel gets divided backwards out of it. $8 million at a $40,000 average deal is 200 deals. At a 20% close rate that's 1,000 opportunities. At 10% conversion from lead, 10,000 leads. Hire accordingly.
Nothing in that sequence is a prediction. It's division.
The number was chosen and the funnel was reverse-engineered to justify it. Had the target been twelve million, every figure downstream would have moved, and not one of them would have been derived from anything a customer did.
The Model Can't Name a Customer
Everyone misses this, and it's the whole thing.
Funnel math is predictable in aggregate and completely unpredictable per deal.
Ask a VP of Sales how much closes this quarter and you get a confident number. Ask which specific accounts, and the answer degrades immediately into a list of names with percentages next to them that nobody in the business actually believes. The forecast holds at the level of the population and dissolves at the level of any individual customer.
That's a strange kind of knowledge to build a business on. You're predicting the behavior of a crowd assembled from people you have never met. It works the way actuarial tables work: sound across ten thousand, useless for the person in front of you.
It also degrades under contact. Every sales organization re-forecasts through the quarter and misses anyway. A prediction you revise three times before it resolves isn't a prediction. It's a plan being defended.
The Quota Is the Confession
If funnel math were genuinely predictive, the revenue would arrive on its own.
It doesn't, so we attach it to someone's compensation. Quota exists because the model doesn't come true by itself. It needs a forcing function, and the forcing function is loss aversion pointed at a human being: hit the number, or the January conversation is the other one.
I have written before that quota is a floor with a party thrown around it. Here's what sits underneath that. A revenue stream that requires a quota in order to materialize isn't predictable. It's coerced, and we have been calling the coercion predictability because the coercion works often enough to pass.
You don't put a quota on things that are actually going to happen.
What You Would Need in Order to Predict Revenue
Set the funnel aside and ask what a real prediction would require.
You would want a buyer who has already bought from you, so willingness to pay is observed rather than assumed. You would want proof you can deliver, so the outcome isn't also a bet on your own execution. You would want to know exactly what they have and what they don't. And you would want a trigger you can see approaching, so that timing is a fact instead of a hope.
That list isn't a wish. It's a description of your existing customer base.
Every item on it is unknown in the funnel and known in the base. The funnel is a population of strangers. The base is a list of named accounts with history attached. One of those is predictable in the ordinary sense of the word, and it isn't the one with the forecast meeting.
Magic Is a Cause You Didn't Observe
So why does expansion feel random?
Because when it happens, nobody can say why it happened. A customer mentions they're adding a team. Someone happens to be on that call. Someone happens to know there's a product for that. The revenue appears without a visible cause, and an effect with no visible cause is the definition of magic.
The cause was there. A customer reached a point where they needed more, which is the most legible event in your business. It went unobserved because nothing was built to observe it.
That's the answer to the whole puzzle. New business isn't more predictable than expansion. It's more instrumented. Stages, conversion rates, a pipeline review every Monday, a named owner, a number that can be missed. Somebody built all of that, deliberately, over years.
Nobody built it for expansion. Then we look at the uninstrumented process, fail to see a pattern, and conclude the revenue is random.
Expansion isn't less predictable than new business. It's less instrumented. Those two claims get confused constantly, and only one of them is true.
What the Instrumented Version Looks Like
It's smaller than people expect.
Customers approaching a milestone, times the share who take the offer attached to that milestone, times the value of that offer. That isn't a model. It's a count.
And it names accounts. Not a population, not a conversion rate applied to strangers, but this customer, this month, this offer, at this price. It's a forecast in a way the funnel never manages, precisely because it can be wrong about something specific.
Which is the point. A number that can be missed is a number somebody will fight for.
The Uncomfortable Part
Expansion isn't predictable in most companies today. That's true, and I'd rather say it plainly than argue around it. An unplanted field is unpredictable too, and for the same reason.
Now the other half of the delivery gate, because this one gets read backwards constantly. Delivery failure precludes expansion. Churn as a number doesn't. Natural attrition precludes nothing, and outgrew-you churn is the strongest expansion signal in the business, since a customer leaving for something you could have sold them is proof the demand was real and unserved. So "we have a lot of churn" isn't a reason to stop. It's a reason to find out which kind you have, because one of those means fix delivery and the other two mean you're already late.
It isn't unpredictable because of anything to do with the revenue itself. It's unpredictable because the four things that would make it predictable haven't been built. Nobody has inventoried what customers could buy next. Nobody has priced those items. Nobody has defined the milestones that signal readiness. And nobody owns the number.
Build those and the randomness goes away, because the randomness was never in your customers. It was in you.
The most predictable revenue in your business is currently the least predicted. That gap isn't a market condition. It's work you haven't done yet.
Where does your company stand? Take the Latent Revenue Test: the six questions, self-served. Ninety seconds, no email required.
Lincoln Murphy formally named and popularized Customer Success starting in 2010 and has spent 15 years connecting it to expansion revenue and commercial outcomes. Read The Premise.